基于最小截取二乘法的道路线形识别算法  

Line type identification method based on least trimmed squares

作　　者：康传利[1] 姚连璧[1] 

机构地区：[1]同济大学测量与国土信息工程系,上海20009

出　　处：《计算机应用》2007年第6期1488-1489,1492,共3页journal of Computer Applications

基　　金：现代工程测量国家测绘局重点实验室开放基金资助项目(ES-SBSM-(06)-09)

摘　　要：根据道路线形的特点,在道路数据采集车获得道路的坐标原始数据的基础上,利用基于Hough变换的直线和圆曲线提取算法对数据进行分组,然后利用基于最小截取二乘法(LTS)的选权迭代法对数据进行平差,从而实现道路线形识别的目的。最后利用条件平差对道路各段线形参数进行平差,得到线形参数的最优化结果。利用本文提出的道路线形识别算法,从原始数据中提取线形参数,进行道路线形识别,在实践中取得了良好的效果。After the positions of mad centerlines were acquired, according to the mad llne types, we first processed the grouping of coordinate data of road centerlines by extracting straight lines algorithm and detecting circle algorithm based on Rough transform, and then carried out adjustment by method for selecting weight Iteration with Least Trimmed Squares （LTS） initial value. Therefore, we reached the goal of road line identification after getting the road line type parameters. At last we adjusted all parameters of line type by using the theory of condition adjustment so that they satisfied a series of objective requirements. This method of road line type identification in this paper has proved its availability by Visual C ＋ ＋ programming finally.

关 键 词：线形识别 HOUGH变换 最小截取二乘法 条件平差 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]